Using Google Trends Data to Assess Reproductive Health Needs in Nigeria During COVID-19

Home , Lockdown (2020)

BIG DATA AND GENDER IN THE AGE OF COVID-19: A BRIEF SERIES FROM UC SAN DIEGO Using Google Trends Data to Assess Reproductive Health Needs in Nigeria During COVID-19

Arnab Dey, Nabamallika Dehingia, Anita Raj Center on Gender Equity and Health, University of California San Diego, USA

Background of data availability around important issues such as family planning. Google Trends data, which reflects searches made Recent research has documented the impact of the on the Google search engine, offers insights into topics of COVID-19 pandemic and consequent lockdowns on access importance for people within a given country, and over time. to key health services, including family planning, in various Given the reduced access to healthcare services, especially countries.1-3 Nigeria is of particular concern, given the around family planning, under the pandemic and related unmet need for contraception4 and high rates of unintended lockdown,1-3 women may be relying more on the internet pregnancies in the country5 prior to the pandemic. Nigeria to access important information on family planning. In also has restrictive abortion laws that impede access except this context, Google Trends data can enable an ecological in cases to save a woman’s life.6 Such restrictive laws can assessment of potentially increasing needs related to result in illegal abortion, which may be unsafe, increasing women’s reproductive health. risk of maternal morbidities and mortality.5,7,8 Studies from other countries indicate declines in contraceptive use and Our Approach increases in unplanned pregnancies and abortions9 due to We utilized the Google Trends Explorer tool11 to download the pandemic and related lockdowns,10 but we continue to and analyze search patterns around pregnancy and abortion lack such data for Nigeria. in Nigeria. We selected the search terms pregnancy and In the absence of direct data on these outcomes, analysis pregnancy test to understand trends in people’s interest in of big data can be useful. Big data platforms such as Google pregnancy, and abortion and misoprostol (a pill used in Trends, Twitter, and other social media are proving highly self-managed medication abortions12,13) for trends in interests beneficial during the COVID-19 pandemic, filling the void in abortion.

Figure 1: Lockdown phases and levels of restrictions implemented in Nigeria during the COVID-19 pandemic • Eased lockdown (Phase 3) • Eased lockdown (Phase 1) • Restrictions were further eased during this period • Nightly curfew between 8pm - 6am • Nightly curfew between 12am - 4am • Neightborhood markets, banks, and government o ces • Federal and State government o ces to operate normal • allowed to operate with limitied sta • working hours. • Interstate travel still prohibited except for essential services • International airspace reopened

Lockdown Lockdown Lockdown Lockdown Phase 1 Phase 2 Phase 3 Phase 4 March 31- May 4- June 2- Sept 4- May 3, 2020 June 1, 2020 Sept 3,2020 Feb 26, 2021

• Strict shelter at home, except for • Eased lockdown (Phase 2) • performing essential services • Nightly curfew between 8pm - 6am • Restriction on interstate travel • Interstate travel and mass gatherings still prohibited • and public commuting • States were allowed to relax restrictions on religious places We specified the geography as Nigeria and the period of smooth data that is characterized by a lot of fluctuations.15 interest as December 1, 2019 and January 31, 2021 on We then used changepoint analysis to identify periods where Google Trends Explorer, and then downloaded the data there was a meaningful change in the interest over time corresponding to interest over time for each search term for each search term. Changepoint analysis is a statistical separately. It is important to note that data pulled from the method that identifies periods in time-series data that are Google Trends Explore tool is normalized to a scale of 0–100 substantially different from neighboring periods.16,17 by dividing each data point by the total searches made for a To understand how these variations were affected by the specific geography and study period.14 Thus the data available COVID-19 pandemic and the lockdowns implemented to us did not indicate the actual number of searches made on in the country, we juxtaposed timelines of key lockdown our search terms but represented a scaled version of those phases in Nigeria on the time series plots. We considered searches made in Nigeria over our study period. four major phases of lockdowns and mapped different levels We first applied a local regression, LOESS, to the downloaded of restrictions implemented in Nigeria during these phases dataset and plotted over-time to understand how search (Figure 1).18-24 patterns varied with time. LOESS is a statistical technique to

Figure 2: Google Search Trends for “pregnancy” (left) and “pregnancy test” (right) in Nigeria between Dec 01, 2019 and Jan 31, 2021 100 100

75 75

50 50 INTEREST OVER TIME INTEREST OVER

25 25 Lockdown Phase 1 Lockdown Phase 1 Lockdown Lockdown Phase 3 Lockdown Phase 3 Lockdown Lockdown Phase 2 Lockdown Phase 2 Lockdown Lockdown Phase 4 Lockdown 0 0 Phase 4 Lockdown Jan 20 Mar 20 May 20 Jul 20 Sep 20 Nov 20 Jan 21 Jan 20 Mar 20 May 20 Jul 20 Sep 20 Nov 20 Jan 21

Figure 3: Google Search Trends for “abortion” (left) and “misoprostol” (right) in Nigeria between Dec 01, 2019 and Jan 31, 2021 100 100

75 75

50 50 INTEREST OVER TIME INTEREST OVER

25 25 Lockdown Phase 1 Lockdown Phase 1 Lockdown Lockdown Phase 3 Lockdown Phase 3 Lockdown Lockdown Phase 2 Lockdown Phase 2 Lockdown 0 Phase 4 Lockdown 0 Phase 4 Lockdown Jan 20 Mar 20 May 20 Jul 20 Sep 20 Nov 20 Jan 21 Jan 20 Mar 20 May 20 Jul 20 Sep 20 Nov 20 Jan 21

BIG DATA AND GENDER IN THE AGE OF COVID-19: A BRIEF SERIES FROM UC SAN DIEGO Results Findings from the Google Trends analysis mapped against the lockdown phases in Nigeria indicate that people’s interest in all four search terms increased at the beginning of the first round of lockdown. Figure 2 shows that searches related to pregnancy peaked during the first phase of the lockdown in Nigeria. Interest in both the terms “pregnancy” and “pregnancy test” peaked and then gradually returned to pre- lockdown levels around June 2020. Figure 3 shows a similar trend in interest over time for searches related to abortion. The search terms “abortion” and “misoprostol” peaked in the first phase of the lockdown (March–April) and then stabilized from June 2020 onwards.

Implications Our analysis shows an increased interest in search terms related to pregnancy and abortion at the onset of the COVID-19 related lockdowns in Nigeria, suggesting that women may have had less control over family planning and greater concern about carrying a pregnancy and having a child during this period. While these findings are not definitive proof of the linkage between the pandemic and restricted access to reproductive care, they do suggest the need for prioritization of sexual and reproductive services, especially for abortion, under lockdowns in Nigeria. In addition, we do not know if the health system was able to respond to this increased demand for abortions during the period. As abortions are legally permitted in Nigeria only to save a woman’s life, it is possible that this increased demand for abortion procedures and medication was met by informal health service providers25 or was self-managed.13 Informal providers may provide misinformation about the medication, its administration, and its possible side-effects,25 putting the health of women at risk. Our analysis also underscores the potential of Google Trends to provide insights related to women’s sexual and reproductive health and rights (SRHR). This approach can easily be replicated with a focus on other countries. This could be particularly useful during situations like the present pandemic in which in-person data collection is challenging. Comparison of Google Trends data with survey data on availability of services and health facilities in sub-regions can also generate insights on demand-supply gaps. These findings call for further research to understand the patterns of SRHR behaviors and need for services during the COVID-19 pandemic across national settings. Such an understanding can help shape policies that meet the SRHR needs of women, particularly in times of crisis.

BIG DATA AND GENDER IN THE AGE OF COVID-19: A BRIEF SERIES FROM UC SAN DIEGO References sellers: a prospective cohort study in Lagos State, Nigeria. BMJ Open 2020; 10(5): e034670. 1. Dasgupta A, Kantorová V, Ueffing P. The impact of the COVID-19 crisis on meeting needs for family planning: a 13. Jayaweera RT, Moseson H, Gerdts C. Misoprostol in global scenario by contraceptive methods used [version the era of COVID-19: a love letter to the original medical 2; peer review: 1 approved, 2 approved with reservations]. abortion pill. Sexual Reproductive Health Matters 2020; Gates Open Research 2020; 4(102). 28(1): 1829406. 2. Ahmed SAKS, Ajisola M, Azeem K, et al. Impact of the 14. Trends Help. FAQ about Google Trends data. https:// societal response to COVID-19 on access to healthcare support.google.com/trends/answer/4365533?hl=en. for non-COVID-19 health issues in slum communities of 15. Cleveland WS, Grosse E, Shyu WM, Chambers JM, Hastie Bangladesh, Kenya, Nigeria and Pakistan: results of pre- TJ. Statistical models in S. Local regression models 1992: COVID and COVID-19 lockdown stakeholder engagements. Chapter 8. BMJ Global Health 2020; 5(8): e003042. 16. Matsa KE, Mitchell A, Stocking G. Methodology. 2017. 3. Endler M, Al-Haidari T, Benedetto C, et al. How the https://www.journalism.org/2017/04/27/google-flint- coronavirus disease 2019 pandemic is impacting sexual and methodology/. reproductive health and rights and response: Results from a 17. Taylor WA. Change-point analysis: a powerful new tool global survey of providers, researchers, and policy-makers. for detecting changes. 2000. Acta Obstetricia et Gynecologica Scandinavica; n/a(n/a). 18. Adebowale N. COVID-19: Nigeria extends PhaseTwo 4. Akamike IC, Okedo-Alex IN, Eze II, Ezeanosike OB, Uneke of eased lockdown by four weeks. 2020. https://www. CJ. Why does uptake of family planning services remain premiumtimesng.com/news/headlines/407194-covid-19- sub-optimal among Nigerian women? A systematic review nigeria-extends-phasetwo-of-eased-lockdown-by-four- of challenges and implications for policy. Contraception and weeks.html. Reproductive Medicine 2020; 5(1): 30. 19. Adebowale N. COVID-19: Nigeria’s third phase of eased 5. Bankole A, Adewole IF, Hussain R, Awolude O, Singh lockdown to last four weeks – Official. 2020. https://www. S, Akinyemi JO. The Incidence of Abortion in Nigeria. Int premiumtimesng.com/news/headlines/412306-covid-19- Perspect Sex Reprod Health 2015; 41(4): 170-81. nigerias-third-phase-of-eased-lockdown-to-last-four- 6. Center for Reproductive Rights. Nigeria’s abortion weeks-official.html. provisions. https://reproductiverights.org/world-abortion- 20. Ajayi O. Phase Two of Eased Lockdown to last for 4 laws/nigerias-abortion-provisions. weeks - FG. 2020. https://www.vanguardngr.com/2020/06/ 7. Akande OW, Adenuga AT, Ejidike IC, Olufosoye AA. phase-two-of-eased-lockdown-to-last-for-4-weeks- Unsafe abortion practices and the law in Nigeria: time for %E2%80%95-fg/. change. Sexual and reproductive health matters 2020; 28(1): 21. Presidential Task on COVID-19. Implementation 1758445. Guidelines for Lockdown Policy. 2020. https://www. 8. WHO. Unsafe abortion Global and regional estimates of statehouse.gov.ng/wp-content/uploads/2020/04/PTF- the incidence of unsafe abortion and associated mortality. COVID-19-Guidance-on-implementation-of-lockdown- Sixth ed; 2008. policy-FINAL.docx-2.pdf. 9. Caruso S, Rapisarda AMC, Minona P, Care RH. Sexual 22. US Dept of State Overseas Security Advisory Council. activity and contraceptive use during social distancing and Health Alert: Nigeria, Guidelines for Eased Lockdown. 2020. self-isolation in the COVID-19 pandemic. The European https://www.osac.gov/Country/Nigeria/Content/Detail/ Journal of Contraception 2020; 25(6): 445-8. Report/047fa10f-1266-4b81-8063-189a5ad4f5ea. 10. Yuksel B, Ozgor F, Obstetrics. Effect of the COVID-19 23. Olisah C. COVID-19: FG extends phase 3 eased lockdown pandemic on female sexual behavior. International Journal by one month. 2021. https://nairametrics.com/2021/01/26/ of Gynecology 2020. covid-19-fg-extends-phase-3-eased-lockdown-by-one- 11. Google News Initiative. Basics of Google Trends. month/ https://newsinitiative.withgoogle.com/training/ 24. Presidential Task on COVID-19. Implementation lesson/5748139575214080?image=trends&tool=Google%20 Guidelines for Phase 3 of eased lockdown. 2020. https:// Trends. statehouse.gov.ng/covid19/wp-content/uploads/2020/09/ 12. Stillman M, Owolabi O, Fatusi AO, et al. Women’s self- F3-Implementation-Guidelines-for-Eased- reported experiences using misoprostol obtained from drug Lockdown_05092020.pdf

BIG DATA AND GENDER IN THE AGE OF COVID-19: A BRIEF SERIES FROM UC SAN DIEGO 25. Guttmacher Institute. Women’s Experiences with Self- Managed Misoprostol Abortions in Lagos State, Nigeria. 2020. https://www.guttmacher.org/fact-sheet/womens- experiences-with-misoprostol-abortions-nigeria#. The code for the analysis can be found here: https://github.com/akdey01/Nigeria-srhr-google-trends

The Center on Gender Equity and Health at UC San Diego’s EMERGE work is supported by the Bill and Melinda Gates Foundation (Grant #s: OPP1163682 and INV-018007.

BIG DATA AND GENDER IN THE AGE OF COVID-19: A BRIEF SERIES FROM UC SAN DIEGO